springload

(Published originally in ClickZ, September 2008) by Eric Picard

It occurs to me that in 10 years, Facebook will know what every ex-girlfriend and ex-boyfriend of an entire generation looks like. They already know what millions of people’s children look like and obviously have numerous images of almost every person that uses its service.

I was talking with a friend the other day about the fact that people haven’t considered the ramifications of Moore’s law (define) on real-time image processing. With more powerful computers and the increases in processing power growing more significant in 10 years, many things we think of today as technically impossible or, at the very least, technically difficult will no longer be. Certainly this will impact technologies like targeting and analytics; it will also impact computer graphics. Looking across both of these worlds and their intersection, it’s easy to start predicting how this could come together.

It won’t be long before the kind of photo and video compositing done painstakingly by hand with lots of CPU horsepower today will be handled in real time on a consumer PC or even on servers in the cloud. This means advertising could be assembled in real time, too. Some companies have been doing this for a while. Visible World, for instance, has enabled creative shops to build template-driven ads that enable elements of the video to be swapped out based on targeting parameters. Near Mother’s Day, residents of an affluent neighborhood might see the expensive flower arrangement while residents of a working-class neighborhood see the inexpensive flower one.

But the kinds of things we’ll see in the next 10 years will make this seem amateurish and quaint. Imagine the following commercials:

• A man stepping out of the new Lexus sedan catches your eye, as he seems somewhat familiar. As he crosses over to the trunk, winking at the attractive (and also somewhat-familiar looking) woman passing through the parking lot, you notice something. He looks an awful lot like you! He opens the trunk and hefts a set of golf clubs. The scene cuts to him beautifully teeing off into the sunrise. You really pay attention — because it’s almost as if someone had peered into your dreams and put them on the television. And you appear in the commercial as an idealized, slightly more chiseled, rugged, and handsome version of yourself.
• A woman who looks oddly like your wife is getting three kids roughly the age, size, and look of your own kids into a minivan that matches the criteria of cars you’ve been shopping for just this week. The eight-year-old boy is carrying a soccer ball — just as your son would be. The toddler even carries a stuffed animal that resembles the one your daughter carries with her everywhere! You see the kids calmly watching a movie on the installed screens, and they seem quite comfortable. The mom seems calm, relieved to have such a nice ride that all the kids enjoy getting into — without squabbling.
• An oddly appealing woman is fly-fishing. She seems so familiar, like you know her from somewhere. The ad focuses in on the graphite rod she’s using, just like the one you were shopping for online last week but didn’t buy. You keep watching because the woman in the ad has such a nostalgic appeal to you. It’s almost as if she were a combination of three women you dated in college. And in truth, she is.

All these commercial seem like science fiction but aren’t far-fetched at all. We think about profile-based targeting as dealing with our habits and anonymously delivering products we’re interested in. But there’s no reason that down the road technology won’t enable the situations I just described. And while the privacy implications are vast — and the ads may seem a bit creepy — over time they may become acceptable. As we’ve seen in numerous studies, the current younger generation has very different feelings about privacy than older generations. And opting in to scenarios like I described may be quite commonplace in 10 years.